So I'm talking about a project which is closely connected to Mark's project because you can see that Mark is one of the co-authors here in this project. And so the idea now is actually to use some larger maps with, in this case, the ERR5 pre-analysis data to come to a more global analysis of the read potential and

then, as a somewhat new result, which we'll for the first time show in here, approaching an estimate of the kind of revenue which one would expect using what price data in addition and we'll try to motivate them. I should say in the beginning that this is a somewhat larger collaborative effort

by now. We started that somewhere after the last SALE and you can see Mark and Rodent as very important members of the team here. I'm Ede Damals from Travis University, the third university contributing and we have a number of additional students who contributed to parts of the project and we have also

everyone in Europe as one of the participating associations and I think this collaborative effort makes it also very interesting. I'll talk a little bit about what we published already and then I come to the uncertainty estimation, that was the important question we should just take, and then we come to the assessment of the resource potential and including some first spot market prices,

some revenue estimates. So here you can see the paper about the resource analysis which was published last time. It's an open access paper and what we did there was using the ERR5 reanalysis data which Mark mentioned on a European map and estimated the wind resource potential

that is without any power characteristics of the urban wind energy device itself included. But looking at the way how the energy would be expected by urban wind energy as compared to conventional wind where at conventional wind you always stay at one altitude range whereas with urban wind energy you have the ability to adapt.

It's not the first study like this, it was also done I think already several years ago for example by a study by Stephanie Mann who used data from Eon from one of the wind parks for that, but we tried to expand it to a relatively large range. The ERR5 reanalysis data set was also already mentioned.

You can find everything which you need to know about it here. It's open data, you can use it for your papers and we also have a excerpt of this data available on a link which you will see at the end of the talk which just concerns the wind because this data obviously includes basically all atmospheric variables

and so it's a little bit large to use everything. Now obviously what we do and just want to reiterate then is something which in this case here and what I used predates the additional study which Mark just introduced so in the future we can obviously build part of what I will show also on these clusters

studied by Mark. Now again what was done for the paper and what was also done for the results here is to look at the time evolution at every given point on this grid on the European map and then we have a certain distribution of wind speed at a given altitude

and you can now select the altitude with the best wind speed within some variations and then look at that given altitude with the best wind speed at the wind speed which you have there and that obviously depends on the profile of the wind speed and this is almost the topic of the previous talk.

What you find in this talk now is a map which is interpolated to a 1 degree times 1 degree grid and it uses the data from 2010 to 2017 for most of the plot. And so the central point is that we find dramatically improved wind at variable heights

if we select the correct height range so that was to be expected I guess. So you can see here for example three different wind profiles for a location in the English channel so a location which is expected to have a very low wind shear and so you see a 100 meter, 500 meter and 1500 meter fixed profile

and then in blue you find the wind profile which you find if you adapt the range to the given wind position at any given time and you can see that at 500 meters here in blue you have essentially the better wind profile than even at 1000 meters fixed and that is just because you can at any given point

use the characteristics of the wind which does not just follow the log profile as you can see. And then of course you can also look at that picture for different ceilings and the interesting thing is that you really move away from basically non-usable wind speeds

and so that obviously should give you as it is to be expected a significant increase in the reliability of the available wind and that really comes with the ability of the altitude and I think that is a very nice case for a low wind and this is what made me so interested about it since mainly I am part of the system.

With that you can make nice plots like that one which you made for a one wind energy a one wind zero brochure where you look at the mean wind speed for average still by half altitudes and compare this on the same color scale with the mean wind speed for variable altitude after 500 meters

and you can see that even over the North Sea you see significant improvements there which just comes from using that variable characteristic of the wind which we have seen so nicely in Mars. Now that is all very nice but obviously we wanted to improve on that and so we thought the following areas might be interesting.

So one thing which is purely technical and therefore unfortunately it is not shown here is that the ERA5 data is organized in model levels which are defined by pressure mostly and so they vary with time so we just have basically fixed pressure levels in this first analysis

and now thanks to Fiona's work we have a kind of time dependent calculation of the exact altitude where the given pressure level is. And the next interesting thing which I will show a little bit here is to develop a reliable estimate of the uncertainty of the ERA5 reanalysis at least for one location

and then obviously we want to have a way to improve the power curve characteristics that we can do in different ways and we will try now to converge on let's say a complex enough but simple enough to use implementation of that so what you see here is a prototype

and even different approaches like what Mark showed before and even what maybe won't come out of our discussions with the whole industry one can come up to a conclusion where people agree that this is a reasonable enough level of precision which is still handleable in terms of computing. And obviously then the next thing is very important for the whole field.

I personally have to say I don't know anything about how much such an energy device will cost once it's really economically usable but what we can already look at is what is the electricity price and what would the different characteristics of the urban wind energy of how it produces the energy and the associated change

in how reliable it is for different wind conditions how would that interact with the variable spot market prices. So that was the impact, the idea of the further developments and so the kind of more global idea which we really want to include the whole community and invite participation is to go towards something like an open tool

for determining the economic impact of urban wind energy using the most precise global wind data which is available. So in principle you can say I select some power characteristics in some reasonably complex way and then I select some location and then I see what I'm looking for.

Okay but first let's come back to the reliability of the reanalysis itself. So we selected freely available LIDAR data in reasonably good shape that's a very important point, which is not in the ERA5 dataset itself included and that is very difficult to find.

And so we find the weight of this LIDAR net measurement at the Wattis time airfield you can see the location here and then you can add the net. Ah very good. And so we see that the global shapes between ERA5 and this measurement after Santina agree perfectly.

So basically 5. So question answer, it's very good. So it's good enough because in principle it just depends on the overall PDF. But of course we want to know how good it is at any given point of time and so we can estimate from the data itself because it's two LIDARs

we can estimate the position of the LIDAR and then we can compare the at any given time the wind speed at different altitudes you see one altitude here between the LIDAR measurement and the ERA5 measurement and then we will radically subtract the optional uncertainty from the Wattis time measurement

and then we find the time resolved uncertainty of the ERA5 reanalysis which is of about 1.3 meters per second up to 1144 meters of altitude and this is a very good position I would say for something which is a global peak basically for an awful lot of measurement and the first 50 data levels is the city over there

it's max. Okay now with all of that together you can again go into the source potential and you can see these funny plots which you can also find in the paper they change a little bit compared to the paper because of these improvements in the altitude calculation and what you can take away is basically from the complex map one thing to look at for example here is the 5.5 percentile

which means there is the wind power available for at least 95% of the time and so you can see the power density here and the ERA5 per square meter but the more interesting thing is to compare the increase factor which you would get at variable harvesting altitude compared to fixed harvesting altitude in this case

so 100 meters you can play the game with different fixed ones but variable is always better and you can see if you follow that line that approximately along the coast lines also here outside of Scotland you can already see a factor 2 improvement of this 5th person passage it will tell you that especially the reliability at rather low wind speeds increases dramatically

and that would be one of the areas where one could look at unique selling points of their own advantage but now for the latest results let's look a little bit at the spot prices because my personal feeling about that is always that we really have to look beyond just full load hours and LCOE

and the reason is simply that the electricity price is not constant and what you can see here you can look at the data yourself here and play with it is that what is plotted here is the residual load on the German grid for example and so this is just defined as the load on the grid minus the wind power minus the solar power

and so you can see this clear correlation which you have here so if the residual load is low so that is basically renewables make up the almost full energy production then prices are zero and can even turn negative and then if there is not so much wind and solar then you have intermediate prices and if there is basically no wind and solar then prices will be taken

so if you manage with double mint energy to produce energy here then this is much, much, much, much more so like around 40 years per megawatt hour than if you produce energy when everyone else produces a lot and so for me this is kind of the key thing which we now would like to connect

now since we didn't have any funds for this project we got spot prices which we got for free and not those which are in here and those which we got for free are Danish and Swedish spot market prices and therefore I will not read with this data but I think the principle should be the same and the next thing obviously you need is a power curve

and again you can use different kinds of simulation so for that we got Elinor Maas from Charles University who is also sitting up there on board and so she did a simulation of a two megawatt wing with six degrees of freedom control with an optimal control optimization of the average power output over cycle

and then using these results just in order that we don't have to do the simulation for every given in condition we scale these results with a kind of simple scaling noise for different tether angles and tether lengths now this is just a technology demonstrator and we as a field would have to determine what kind of complexity you would have with Elinor Maas

and let me compare this to a recovery turbine and the characteristics of that you can find here then since it's Swedish spot market prices are just randomly and this is just, there is no optimization here so this is just something I made to be ready in time and so I just selected one spot in the middle between Leuven here

and this spot here, so between Sweden and Germany in the Baltic Sea and I just selected two points of time out of which I will show a random one and this is the beginning of June in 2017 and what you see there now is you can now really follow we have a mid energy device on its path

through the time of the first week of June 2017 and you can see that obviously it very often refers to relatively low harvesting altitudes an average of 100 meter at the bottom it's not allowed to go lower but then as the wind speed drops significantly it ventures up where at that point of time there is more wind

at some time obviously it can happen that there is so little wind that it doesn't help anymore and then it goes down obviously to the lower area so this is a very extreme case I guess and over land you would definitely see a much different characteristic that we already know but this is just one of the two arbitrarily chosen time points at the only location which I can show you up to now

and as of already there we see very interesting things because now we can compare the turbine with the hub at 100 meter which is a little bit overestimated for that specific model I think it was maybe something but gives you approximately the right dimensions and here is an able wind energy device with a ceiling of 500 meter

and now we are looking here at the power which is produced and you can see this is already much different for the able wind energy device than for the turbine simply because even if you don't vary the harvesting altitude you can vary the angle of the tether and the length

and therefore adjust the shape of your power curve with the scaling loss which we included to the wind conditions even that helps in addition you then have the possibility to vary the harvesting altitude and then you can put in the spot market prices and you get the income which you will get from the turbine relative to the income which you will get from the able wind energy device

and so even if you don't know the price of the able wind energy device you can at least now say you get more wind So that's good I hope someone in time brings me to the summary so we can obviously do each part of this analysis as I said with the input of everybody which is very relevant to really come to a framework

where we can put in these power curves into the system in a way that the people agree first that the power curves which the competitors put in are reliable so this is the most important thing and such that people agree that the level of complexity which might be a little bit low on that one but I think it's already giving a relatively interesting order of magnitude

and that this is really sufficient and obviously then we want to as well get more complete results than what I could show you because that is what we're developing at the moment So I think we can conclude that with this ERA-5 reanalysis data set from the European Centre for Medium-Terrebellar Forecast

we have a very precise data set available which is with a 31 by 31 kilometre grid in its finest spacing relatively fine-grained and gives an extremely good basis for such studies simply because we now know at least from one location in Great Britain

that it's also very reliable So that allows us then to study the resource potential including the power curve with the form for mid-data and we found a significant revenue increase due to the increased capacity factor and the adaptability of the power curve characteristics

to the wind characteristics and so if you just take that last plot here and just integrate over the plot to find 3,976 euro for the turbine and 6,400 euros for all the other medium energy devices that might be off by 10% but it's still good

So obviously we will try to get out much more interesting information from that expand it over the map and publish it again but until then you can already play with the code because everything is open source Not all the new functionality which Mark and I now introduced is already in the public code

but the previous state you can find here in this repository and the data associated with that you can find here and everybody is invited to contact us Thank you very much